2 Implementation of the 'import' subcommand.
5 from pathlib import Path
9 from nominatim.db.connection import connect
10 from nominatim.db import status, properties
11 from nominatim.version import NOMINATIM_VERSION
12 from nominatim.errors import UsageError
14 # Do not repeat documentation of subcommand classes.
15 # pylint: disable=C0111
16 # Using non-top-level imports to avoid eventually unused imports.
17 # pylint: disable=E0012,C0415
19 LOG = logging.getLogger()
23 Create a new Nominatim database from an OSM file.
28 group_name = parser.add_argument_group('Required arguments')
29 group = group_name.add_mutually_exclusive_group(required=True)
30 group.add_argument('--osm-file', metavar='FILE',
31 help='OSM file to be imported.')
32 group.add_argument('--continue', dest='continue_at',
33 choices=['load-data', 'indexing', 'db-postprocess'],
34 help='Continue an import that was interrupted')
35 group = parser.add_argument_group('Optional arguments')
36 group.add_argument('--osm2pgsql-cache', metavar='SIZE', type=int,
37 help='Size of cache to be used by osm2pgsql (in MB)')
38 group.add_argument('--reverse-only', action='store_true',
39 help='Do not create tables and indexes for searching')
40 group.add_argument('--no-partitions', action='store_true',
41 help=("Do not partition search indices "
42 "(speeds up import of single country extracts)"))
43 group.add_argument('--no-updates', action='store_true',
44 help="Do not keep tables that are only needed for "
45 "updating the database later")
46 group = parser.add_argument_group('Expert options')
47 group.add_argument('--ignore-errors', action='store_true',
48 help='Continue import even when errors in SQL are present')
49 group.add_argument('--index-noanalyse', action='store_true',
50 help='Do not perform analyse operations during index')
54 def run(args): # pylint: disable=too-many-statements
55 from ..tools import database_import
56 from ..tools import refresh
57 from ..indexer.indexer import Indexer
58 from ..tools import postcodes
59 from ..tokenizer import factory as tokenizer_factory
61 if args.osm_file and not Path(args.osm_file).is_file():
62 LOG.fatal("OSM file '%s' does not exist.", args.osm_file)
63 raise UsageError('Cannot access file.')
65 if args.continue_at is None:
66 database_import.setup_database_skeleton(args.config.get_libpq_dsn(),
69 rouser=args.config.DATABASE_WEBUSER)
71 LOG.warning('Installing database module')
72 with connect(args.config.get_libpq_dsn()) as conn:
73 database_import.install_module(args.module_dir, args.project_dir,
74 args.config.DATABASE_MODULE_PATH,
77 LOG.warning('Importing OSM data file')
78 database_import.import_osm_data(Path(args.osm_file),
79 args.osm2pgsql_options(0, 1),
81 ignore_errors=args.ignore_errors)
83 with connect(args.config.get_libpq_dsn()) as conn:
84 LOG.warning('Create functions (1st pass)')
85 refresh.create_functions(conn, args.config, False, False)
86 LOG.warning('Create tables')
87 database_import.create_tables(conn, args.config,
88 reverse_only=args.reverse_only)
89 refresh.load_address_levels_from_file(conn, Path(args.config.ADDRESS_LEVEL_CONFIG))
90 LOG.warning('Create functions (2nd pass)')
91 refresh.create_functions(conn, args.config, False, False)
92 LOG.warning('Create table triggers')
93 database_import.create_table_triggers(conn, args.config)
94 LOG.warning('Create partition tables')
95 database_import.create_partition_tables(conn, args.config)
96 LOG.warning('Create functions (3rd pass)')
97 refresh.create_functions(conn, args.config, False, False)
99 LOG.warning('Importing wikipedia importance data')
100 data_path = Path(args.config.WIKIPEDIA_DATA_PATH or args.project_dir)
101 if refresh.import_wikipedia_articles(args.config.get_libpq_dsn(),
103 LOG.error('Wikipedia importance dump file not found. '
104 'Will be using default importances.')
106 if args.continue_at is None or args.continue_at == 'load-data':
107 LOG.warning('Initialise tables')
108 with connect(args.config.get_libpq_dsn()) as conn:
109 database_import.truncate_data_tables(conn, args.config.MAX_WORD_FREQUENCY)
111 LOG.warning('Load data into placex table')
112 database_import.load_data(args.config.get_libpq_dsn(),
114 args.threads or psutil.cpu_count() or 1)
116 LOG.warning("Setting up tokenizer")
117 tokenizer = tokenizer_factory.create_tokenizer(args.config)
119 if args.continue_at is None or args.continue_at == 'load-data':
120 LOG.warning('Calculate postcodes')
121 postcodes.import_postcodes(args.config.get_libpq_dsn(), args.project_dir)
123 if args.continue_at is None or args.continue_at in ('load-data', 'indexing'):
124 if args.continue_at is not None and args.continue_at != 'load-data':
125 with connect(args.config.get_libpq_dsn()) as conn:
126 SetupAll._create_pending_index(conn, args.config.TABLESPACE_ADDRESS_INDEX)
127 LOG.warning('Indexing places')
128 indexer = Indexer(args.config.get_libpq_dsn(),
129 args.threads or psutil.cpu_count() or 1)
130 indexer.index_full(analyse=not args.index_noanalyse)
132 LOG.warning('Post-process tables')
133 with connect(args.config.get_libpq_dsn()) as conn:
134 database_import.create_search_indices(conn, args.config,
135 drop=args.no_updates)
136 LOG.warning('Create search index for default country names.')
137 database_import.create_country_names(conn, args.config)
139 webdir = args.project_dir / 'website'
140 LOG.warning('Setup website at %s', webdir)
141 refresh.setup_website(webdir, args.config)
143 with connect(args.config.get_libpq_dsn()) as conn:
145 dbdate = status.compute_database_date(conn)
146 status.set_status(conn, dbdate)
147 LOG.info('Database is at %s.', dbdate)
148 except Exception as exc: # pylint: disable=broad-except
149 LOG.error('Cannot determine date of database: %s', exc)
151 properties.set_property(conn, 'database_version',
152 '{0[0]}.{0[1]}.{0[2]}-{0[3]}'.format(NOMINATIM_VERSION))
158 def _create_pending_index(conn, tablespace):
159 """ Add a supporting index for finding places still to be indexed.
161 This index is normally created at the end of the import process
162 for later updates. When indexing was partially done, then this
163 index can greatly improve speed going through already indexed data.
165 if conn.index_exists('idx_placex_pendingsector'):
168 with conn.cursor() as cur:
169 LOG.warning('Creating support index')
171 tablespace = 'TABLESPACE ' + tablespace
172 cur.execute("""CREATE INDEX idx_placex_pendingsector
173 ON placex USING BTREE (rank_address,geometry_sector)
174 {} WHERE indexed_status > 0
175 """.format(tablespace))